Cancer patients want to know what caused their cancer. Pragmatically, the causes of cancer in an individual can be broadly classified into Figure 9 Relationship between relative risk and risk factor prevalence on the population attributable risk.
environmental (cumulative exposure over a lifetime to a variety of carcinogenic and protective factors), genetic, and spontaneous (51).
1. Spontaneous. Often overlooked are the spontaneous causes of can-cer. By spontaneous, what is meant is that a certain amount of cancer is due to ‘‘spontaneous’’ or ‘‘background’’ mutation rates (51). These muta-tions generally show a different pattern of DNA lesions compared to those induced by carcinogens. The exact causes of these mutations are not known, but are likely due to things such as background cosmic radiation and body temperature, and reflect the instability of DNA as a result of oxidative damage and other cellular processes. These factors would be expected to show little or no variability in terms of geographical distribution, and thus there will always be a certain background level of cancer in any population.
Doll and Peto (9) have also termed this ‘‘chance,’’ or more simply good or bad luck. At the level of the individual, spontaneous causes of cancer may play an important explanatory role. Knudson (51) has estimated that approximately 15% of cancer may be explained by spontaneous factors.
2. Genetic. Cancer has long been known to aggregate in families, strongly supporting a hereditary component for a certain portion of cancer.
Familial cancers are generally characterized by early age at onset, bilateral tumors in paired organs, multiple primary foci within an organ, distinctive pathology, and often prominent physical findings. They may also be a part of a syndrome that includes multiple cancer sites and=or other diseases (e.g., Von Hippel–Lindau disease and renal cell carcinoma), and are caused by germline (i.e., changes in the constitutional DNA) alterations in single genes that often follow Mendelian patterns of inheritance (i.e. ‘‘major genes’’). However, it must be kept in mind that cancer is a relatively com-mon disease, and thus many persons will have a positive family history of cancer, and some cancer will cluster in families by chance alone. In addition, families often share similar environmental exposures (including residence, diet, and so forth) and this may explain some clustering of cancer within families. Thus, ‘‘familial’’ is not synonymous with ‘‘genetic.’’ Geneticists and genetic epidemiologists use family studies to evaluate the relative con-tribution of genes vs. environment, and to identify new cancer genes.
In contrast to cancer caused by major genes, a second class of genes, often termed ‘‘susceptibility’’ genes (52) are also likely to be important in cancer causation. Susceptibility genes are common variants (polymor-phisms) of genes generally involved in the metabolic activation and detoxi-fication of carcinogens, but they can be involved in other pathways relevant to carcinogenesis including DNA repair. This concept is highly influenced by the field of pharmacogenetics, and examples of this approach include stu-dies of lung cancer and debrisoquine metabolism and GST-mu deficient phenotype and bladder cancer.
An important distinction between these two types of genetic causes of cancer are that while major genes carry a high absolute and relative risk,
they are uncommon in the population and thus have a low population attri-butable risk. In contrast, susceptibility genes are associated with a low abso-lute and relative risk of cancer, but because many of these variants are common in the population, they may have a high population attributable risk. Another important distinction is that the major genes are expected to be less influenced by environmental exposures relative to the suscepti-bility genes, which primarily influence host response to the environment.
Understanding hereditary cancer is expected to give mechanistic insight into the causes of sporadic (i.e., nonhereditary) cancer. However, in the popula-tion, only about 5% of cancer is thought to be due to purely genetic (major gene) causes (51).
3. Environmental, lifestyle, and behavioral factors. A conclusion from the descriptive and analytical epidemiology of cancer is that cancer should be largely, although not completely, preventable and that environmental and behavioral factors should account for a large percentage of the total cases, often estimated at up to 80% of cancer (51). Doll and Peto (9,53) ori-ginally published their estimates of cancer deaths attributable to various environmental and behavioral factors in western populations in 1981, and recently updated this in 1996. As shown in Table 6, tobacco and diet are
Table 6 Estimates of the Proportion of Cancer Deaths Attributable to Environmental and Lifestyle Factors in Western Countries
Factor(s)
Best estimate of proportion (%)
Range of acceptable estimates (%)
Tobacco 33 25–40
Diet 30 20–60
Infection 9 5–15
Hormones 7 5–10
Ionizing radiation 4 2–6
Background 3.5
Medical procedures 0.5
Industry <0.1
Alcohol 3 2–4
Occupation 3 2–4
Pollution <2 <1–2
Atmospheric <1
Water <1
Ultraviolet light 1 0.5–1
Industrial products <1 <1–2
Medical drugs <1 <1–2
Food additives <1 2–1
Other and unknown ? ?
Source: From Ref. 9. Copyright 1996 Oxford University Press.
thought to be the most important causes of cancer, although clearly the ulti-mate role of diet in the etiology of cancer is still being unraveled. The next most important group of factors are infection, hormones, background ioniz-ing radiation, occupation, and alcohol. Of relatively less importance are ultraviolet radiation, industrial products, water and air pollution, and food additives, exposures that tend to receive a disproportionate amount of media attention.
One limitation in interpreting Table 6 is that it does not take into account interactions between exposures. For example, smoking increases the risk of lung cancer, as does exposure to asbestos; however, the risk of lung cancer in smokers exposed to asbestos is much greater than expected based on each risk factor considered individually. Other well-established interactions include smoking and radon for lung cancer, smoking and alco-hol for esophageal cancer, and hepatitis B infection and exposure to afla-toxin for liver cancer. Other interactions are likely.
4. Gene–environment interaction. While carcinogen exposure triggers the onset of cancer, a person’s genetic makeup determines how they respond to the exposure. Thus, genes may increase or decrease risk from the expo-sure and so this is considered a gene–environment interaction. There is cur-rently great interest in identifying interactions between genetic and environmental causes of cancer. As alluded to above, most (but by no means all) of the interest is focused on the interaction of susceptibility genes with environmental exposures. Although unknown at this time, much of the 80% of cancer thought to be due to environmental causes may be due to gene–environment interactions (39). The study of gene–environment interactions will consume much of epidemiological research over the next decade.